We deal with this sort of thing a couple of times a year because we have so many delay-bound applications. Your problem could be due to any one of a great number of causes. Here are just a few to think about...
Electromagnetic signals (electrical or light) travel through physical media at about 2/3 speed of light. This translates to about 1 millisecond per 100 miles. Also, each comm node that a circuit traverses adds delay due to buffers. Some carriers use DACS (digital access and cross-connect systems) extensively and some do not. DACS are full of buffers. So when you add up delay, there can be great disparity between two different carriers. Although unlikely, one of your carriers could even be using SATCOM between the US and Mexico.
Having said all of that, if you do the math, the difference between the two delay times (over 100 ms) would require 10,000 additional terrestrial miles (if considering propagation delay only). Not likely but certainly within the realm of possibility. I have seen circuits go from Colorado, out to California, down to Albuquerque, down to Texas, and ramble up and down the East Coast before getting to D.C. Carriers don't always use the most efficient routes in terms of distance - they use the most efficient route in terms of available bandwidth within their own networks.
Also, bear in mind that there is never just one carrier involved. You may have one as a central POC. But due to government regs, you will typically have one or more long-haul carriers, and at least two local exchange carriers (LECs).
Finally, getting specifics can be very difficult. Carriers consider such internal info to be proprietary and usually aren't very forthcoming.
Here is what we do to solve the problem: when we order a circuit, we specify that the end-to-end delay cannot exceed 100 ms. If it goes beyond that, it is considered an outage and we don't pay. That goes a long way to ensure a good, efficient path.
Good luck
